------- Comment #7 from rguenther at suse dot de 2009-01-26 16:07 ------- Subject: Re: [4.3/4.4 Regression] ICE in set_value_range, at tree-vrp.c:398
On Mon, 26 Jan 2009, jakub at gcc dot gnu dot org wrote: > ------- Comment #6 from jakub at gcc dot gnu dot org 2009-01-26 16:02 ------- > That brings us back to http://gcc.gnu.org/ml/gcc-patches/2008-11/msg00532.html > If the gimplifier shouldn't drop TREE_OVERFLOW bits from the IL, then it is > valid to have TREE_OVERFLOWs set and tree-vrp.c needs to deal with it (use > VARYING range or something similar). Not exactly. Your patch was correct in principle but exposed latent bugs elsewhere. Difficult ones, because fold is shared between FEs and the middle-end and fold and the FEs _do_ care about TREE_OVERFLOW. ... So indeed, working around these issues where they pop up is the way to go in the near future. (this doesn't mean a TREE_OVERFLOW in the IL is valid ...) Richard. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38934